CN117082020B

CN117082020B - Automatic forwarding method for friend circle based on instant messaging software

Info

Publication number: CN117082020B
Application number: CN202311327305.9A
Authority: CN
Inventors: 唐宏才; 阳纯正; 谢玉明
Original assignee: Hunan Shuanghong Technology Co ltd
Current assignee: Hunan Shuanghong Technology Co ltd
Priority date: 2023-10-13
Filing date: 2023-10-13
Publication date: 2023-12-29
Anticipated expiration: 2043-10-13
Also published as: CN117082020A

Abstract

The invention relates to the technical field of data processing, in particular to an automatic forwarding method for a friend circle based on instant messaging software. The method comprises the following steps: acquiring terminal equipment authority of a user; acquiring relevant data of instant messaging software according to the authority of the terminal equipment, and generating browsing interest relevant data; extracting preferred and alternative browsing data from browsing interest related data to generate preferred browsing data and alternative browsing data respectively; the user-defined selection of the alternative browsing data and the preferred browsing data are subjected to data integration to generate pre-forwarding data; optimizing forwarding data content of the pre-forwarding data to generate optimized pre-forwarding data; and constructing an automatic forwarding script according to the optimized pre-forwarding data and a preset automatic script frame, and generating automatic forwarding content data when executing the automatic forwarding script. The invention realizes a more personalized and intelligent automatic forwarding method of the friend circle.

Description

Automatic forwarding method for friend circle based on instant messaging software

Technical Field

The invention relates to the technical field of data processing, in particular to an automatic forwarding method for a friend circle based on instant messaging software.

Background

In the development process of instant messaging software, people are gradually familiar with the instant messaging software, and people want to select forwarding for interested friends, so that more people's attention is harvested. However, the manual forwarding is sometimes time-consuming and laborious, and the complex forwarding process gradually loses interest in forwarding the friend circle, so that the automatic forwarding method for the friend circle can help us solve the problems. However, the automatic forwarding method of the friend circle of the traditional instant messaging software simply forwards the content, and forwards too much content, so that other users do not want to view the forwarded content of the friend circle, and the other users lose interest in browsing the forwarded friend circle.

Disclosure of Invention

Based on the above, the invention provides an automatic forwarding method of a friend circle based on instant messaging software, so as to solve at least one of the above technical problems.

In order to achieve the above purpose, the automatic forwarding method of the friend circle based on the instant messaging software comprises the following steps:

step S1: acquiring terminal equipment authority of a user; acquiring relevant data of instant messaging software according to the authority of the terminal equipment, and generating browsing interest relevant data;

Step S2: extracting preferred and alternative browsing data from browsing interest related data to generate preferred browsing data and alternative browsing data respectively;

step S3: sequencing the priority of the browsing data of the preferred browsing data to generate the priority browsing data, and integrating the selected alternative browsing data with the priority browsing data by self definition to generate pre-forwarding data;

step S4: optimizing forwarding data content of the pre-forwarding data to generate optimized pre-forwarding data;

step S5: and constructing an automatic forwarding script according to the optimized pre-forwarding data and a preset automatic script frame, generating automatic forwarding content data when executing the automatic forwarding script, and collecting a log file of the automatic forwarding content data to feed back to the terminal.

According to the invention, the browsing data of the instant messaging software is monitored in real time by acquiring the authority of the terminal equipment of the user, and the interest degree of the browsing data of the user is calculated. This enables the generation of highly personalized browsing interest related data. The preferred and alternative data extraction of browsing interest related data enables a screening and sorting of content, which is beneficial for the user experience, since it ensures that after forwarding a circle of friends, other users see the most relevant and interesting content (preferred browsing data) first, thereby improving satisfaction and interaction rate. Meanwhile, the alternative browsing data provides various choices for users, promotes wider content transmission on the platform, increases the attraction of the platform, and provides a more effective data basis for personalized recommendation. By prioritizing the browsing interest levels associated with the preferred browsing data, highly personalized content recommendations can be provided to the user, ensuring that they see the most interesting content first, improving user engagement and satisfaction, and increasing user residence time on the platform. Meanwhile, the alternative browsing data and the priority browsing data are selected by user definition to integrate the data, so that the diversity and the attraction of the content are further enhanced, the requirements of different users are met, the viscosity of a platform and the interaction of the users are improved, and a more accurate data base is provided for content recommendation. By optimizing the text content and the image content of the pre-forwarding data, more attractive and quality content can be provided for users, the user experience is improved, the readability and visual appeal of the content are improved, and the clicking and participation rate of the users on the content is increased. Through carefully optimizing the forwarded content, the method not only can better meet the interests and expectations of users, but also can improve the propagation effect of the content and promote social interaction among users, and simultaneously provides a medium with more attractive and effective advertising, and is very important for improving the user retention rate and the competitiveness of a platform. By constructing an automatic forwarding script according to the optimized pre-forwarding data and the automatic script framework, efficient and accurate content forwarding is realized, and the user participation is improved, because the user can easily share personalized and attractive content without manual operation, the liveness of the platform is improved. The method and the system are beneficial to providing useful information for further improvement and optimization, improving the efficiency of automatic forwarding, enhancing the competitiveness of a platform and simultaneously providing a data feedback mechanism to ensure that the system is continuously improved and meets the demands of users. Therefore, the automatic forwarding method of the friend circle based on the instant messaging software does not simply forward the content, but optimizes the forwarded content, so that the forwarded content attracts other users more, and the forwarded content is more interested than the original content.

Preferably, step S1 comprises the steps of:

step S11: acquiring terminal equipment authority of a user;

step S12: real-time monitoring of browsing data of the instant messaging software is carried out according to the authority of the terminal equipment, and browsing data is generated;

step S13: performing time sequence extraction on the browsing data to generate time sequence data;

step S14: calculating the interest level of the browsing data by using a browsing interest level calculation formula to generate a browsing interest level;

step S15: and integrating the browsing data and the browsing interest level according to the time series data to obtain corresponding time series data, and generating browsing interest related data.

The invention ensures that the system can legally access the user terminal equipment, provides necessary rights for subsequent operation, properly protects the data privacy and safety of the user, and simultaneously provides legal compliance and operability for data acquisition and analysis. The real-time monitoring of the browsing data of the instant messaging software is carried out according to the authority of the terminal equipment, so that the real-time data monitoring of the instant messaging software is realized, the browsing data of a user is obtained, personalized content is provided, the user experience is improved, and a data basis is provided for the follow-up forwarding of the browsed friend circle data. The extraction of the time series data is helpful to analyze the behavior trend and interest change of the user, and can help the system to better understand the interest of the user in different time periods so as to recommend the content more accurately, for example, the length of time for the user to browse a friend circle in a certain time period is convenient to judge whether the user is interested in the category of the friend circle according to the length. By calculating the interest level of the browsing data, weights can be distributed to different contents, so that the content forwarding is more accurate and effective when the content forwarding is executed for the subsequent user, the browsing data and the browsing interest level are corresponding to the time sequence data, and a detailed user interest map can be created. This map may help better understand the user's interest changes, providing the user with the most relevant content at different points in time, and such dynamic content recommendation helps to enhance the user's viscosity, as they may feel the system more aware of their needs.

Preferably, the browsing interest level calculation formula is as follows:

；

in the method, in the process of the invention,expressed as browsing interest level->Expressed as user browsing end time->Expressed as user browsing start time->Expressed as the ratio of the category to which the content belongs in all categories,/for>Expressed as a proportion of the category to which the content belongs in the category the user browses>Expressed as the ratio of the category to which the content belongs in the user history forwarding content category, +.>Expressed as browsing time corresponding to browsing data, +.>Content history praise number expressed as correspondence of browsing data,/->The number of praise on the same day indicated as the content corresponding to the browsing data,/->Content distribution time indicated as browsing data correspondence, < >>Expressed as an exponential constant, +.>An anomaly adjustment value expressed as a browsing interest level.

The invention utilizes a browsing interest level calculation formula which fully considers the browsing ending time of the userUser browsing start time->The ratio of the category to which the content belongs in all categories->Content belonging category is in the user browsing category +.>Content category ratio in user history forwarding content category>Browsing time corresponding to browsing data>Content history praise number corresponding to browsing data >Content corresponding to browsing data like number of praise on the same day->Content distribution time corresponding to browsing data>Index constant->And interactions between functions to form a functional relationship:

that is to say,the functional relation calculates the degree of interest of the user in browsing the data content by considering the browsing time, praise number, comment number and the like of the user for a certain piece of friend circle content. The user browsing end time and the starting time represent the browsing time length of the user, and the browsing time is longer to drive the browsing interests of the user; the proportion of the category of content in all categories, the proportion of the category of content in the category of user browsing and the proportion of the category of content in the category of user historic forwarding content means that the user has different degrees of interest in different categories of content, for example, the user frequently browses the content of the technology category and their historic forwarding is also dominated by the technology related content, then their interest level may be higher on the content of that category. The history number of praise of the content corresponding to the browsing data and the praise number of the content on the same day represent the influence of the praise number, and the praise number can be regarded as an index of the popularity of the content, if a content obtains a large number of praise in the past, the content can be considered to be more attractive, the interest level of the user can be higher, and the praise number on the same day can be used for considering the real-time popularity of the content; the content distribution time corresponding to the browsing data is used to consider the freshness of the content, and for some fields, the newly distributed content may be more attractive to users, so they may get a higher interest level in the calculation. The functional relation aims at accurately calculating the interest level of the user to the specific content according to the browsing behavior of the user, the characteristics of the content and the historical data, so that personalized content recommendation can be better provided for the user, the user experience is improved, and the attraction of the content is improved. Abnormality adjustment value using browsing interest level +. >Adjusting and correcting the functional relation to reduce abnormal data or errorError influence caused by the difference item, thereby generating browsing interest level more accurately>The accuracy and the reliability of interest level calculation of browsing contents on browsing data are improved. Meanwhile, the weight information and the adjustment value in the formula can be adjusted according to actual conditions and are applied to different browsing data for calculation, so that the flexibility and applicability of the algorithm are improved.

Preferably, step S2 comprises the steps of:

step S21: judging browsing interests of the browsing interest related data, and marking the browsing data of the browsing interest related data as preferable browsing data when the browsing interest level of the browsing interest related data is larger than a preset first interest threshold value;

step S22: judging browsing interests of the browsing interest related data, and marking the browsing data of the browsing interest related data as alternative browsing data when the browsing interest level of the browsing interest related data is not greater than a preset first interest threshold value and is greater than a preset second interest threshold value, wherein the first interest threshold value is greater than the second interest threshold value;

step S23: and judging the browsing interest of the browsing interest related data, and rejecting the browsing data of the browsing interest related data when the browsing interest level of the browsing interest related data is not greater than a preset second interest threshold value.

According to the invention, through comparing the browsing interest level of the browsing interest related data with the preset first interest threshold value, the content of most interest can be accurately determined and marked as the preferable browsing data, which is beneficial to providing highly relevant and deeply personalized content for users, thereby increasing the satisfaction degree and interaction of the users to the friend circle content and promoting the wide spread of the content. The data of the browsing interest level of the browsing interest related data between the first interest threshold value and the second interest threshold value is marked as alternative browsing data, which is beneficial to providing the selection of the diversity of users, meeting the users of different levels of interests, promoting the diversity and the wide spread of the content, and the existence of the alternative data can be more selection of the content forwarding. By comparing the browsing interest level of the browsing interest related data with the second interest threshold, data which is not interested in the user can be eliminated, the quality of the content is improved, the possibility that the user sees the irrelevant or uninteresting content is reduced, the quality of user experience is improved, and meanwhile, the bandwidth and storage resources are saved.

Preferably, step S3 comprises the steps of:

Step S31: carrying out self-defined selection on the alternative browsing data to obtain selected browsing data;

step S32: and sorting the priority of the browsing data according to the browsing interest level related to the preferred browsing data, generating priority browsing data, and integrating the priority browsing data with the selected browsing data to generate pre-forwarding data.

According to the invention, the alternative browsing data is selected by self-definition according to the specific standard, so as to obtain the content which best meets the interests and platform targets of the user, provide highly personalized content, ensure that the user sees the content which the user is interested in, improve the user experience and satisfaction, and can be used for controlling the diversity of the content more accurately so as to meet the requirements of different users. By sorting the priority of the browsing data according to the browsing interest level of the preferred browsing data, the user is ensured to see the most relevant and interesting content first, so that the user satisfaction degree and interactivity are improved, the priority browsing data is integrated with the browsing data selected by user definition, the diversity and attraction of the content are further enhanced, the requirements of different friend circle contents are met, the viscosity of a platform and the user interaction are improved, the accuracy and diversity of personalized recommendation are emphasized, the user is facilitated to participate in the platform content deeply, and the retention rate of the user is improved.

Preferably, step S4 comprises the steps of:

step S41: dividing data of text and image of the pre-forwarding data to generate pre-forwarding text data and pre-forwarding image respectively;

step S42: performing text abstract analysis on the pre-forwarded text data to generate text abstract data;

step S43: data integration is carried out on the text abstract data and the pre-forwarding text data, and optimized text data is generated;

step S44: filtering the pre-forwarded image by using Gaussian filtering to generate a filtered image;

step S45: carrying out local dynamic image optimization processing on the filtered image to generate an optimized image;

step S46: and integrating the optimized text data and the optimized image data of the corresponding sequence to generate optimized pre-forwarding data.

According to the invention, the pre-forwarding data is divided into the text and the image, so that the text and the image content can be optimized respectively, the best quality of the text and the image content in vision and language can be ensured, the attraction of a user to the content can be improved, the user experience can be improved, and the click rate and the interaction of the user can be increased. By performing text abstract parsing on the pre-forwarded text data, more refined and readable text abstract data can be generated, which is beneficial to providing faster content understanding and browsing for users, reducing the possibility of information overload and improving user experience. Integrating the text abstract data with the original text data to generate optimized text data is beneficial to providing richer but not less concise content, and can improve the interest and attraction of the user to the content, thereby increasing the participation of the user. The Gaussian filtering is used for processing the image, so that noise in the image can be removed, the definition and quality of the image are improved, the image content with visual appeal is provided, and the visual experience of a user is improved. And the filtered image is subjected to local dynamic optimization, so that the visual attraction of the image is further improved, the content of the dynamic blurred image is displayed more clearly, and the retention rate and interactivity of a user are improved. The optimized text data and the image are integrated into a whole to generate optimized pre-forwarding data, so that coordination and consistency of text and image contents are ensured, more comprehensive and attractive contents are provided, and interest and participation of users in the contents are improved.

Preferably, step S42 comprises the steps of:

extracting key fields of the pre-forwarding text data to generate key text data;

acquiring text-summary data;

establishing a mapping relation between text content and a text abstract by using a cyclic neural network, and generating an initial abstract prediction model;

model training is carried out on the initial abstract prediction model by using the text-abstract data, and an abstract prediction model is generated;

and transmitting the key text data to a abstract prediction model to perform text abstract prediction, and generating text abstract data.

The invention can capture the most important and relevant information in the text by extracting the key field of the pre-forwarded text data, thereby improving the accuracy and efficiency of the text abstract, being beneficial to a user to quickly acquire the core content of the article, reducing the risk of information overload and improving the user experience. The text-summary data is obtained in order to build a training data set required for training a summary prediction model, which is beneficial to learning the rule of generating the summary from a large number of text examples so as to improve the capability of automatically generating the text summary. By using deep learning models such as a Recurrent Neural Network (RNN), complex mapping relations between text contents and text summaries can be learned, more accurate and coherent text summaries are generated, and the information integrity of the contents is ensured. By using text-abstract data to carry out model training on an initial abstract prediction model, the performance of the model is continuously improved, the generation quality of the text abstract is improved, abstract content with more grammar correctness and consistency is generated, and the understanding and participation of users on the content are improved. The key text data is transmitted to the abstract prediction model to conduct text abstract prediction, so that a refined text abstract can be generated quickly and automatically, a user can be provided with a more efficient content browsing and understanding mode, and the satisfaction degree of the user is improved, because the user can acquire information more quickly.

Preferably, step S45 comprises the steps of:

extracting a local dynamic region of the filtered image to generate a local dynamic image;

carrying out gray level image conversion on the local dynamic image to generate a local gray level image;

image texture extraction is carried out on the local gray level image by using the gray level co-occurrence matrix, and image texture data are generated;

carrying out gray value calculation on the local gray image by using a dynamic image gray value calculation formula to generate a restoration gray value of the local gray image;

carrying out restoration mapping processing on the local dynamic image on the image texture data and the restoration gray value to generate a restoration local dynamic image;

and carrying out local dynamic image optimization on the filtered image according to the repaired local dynamic image to generate an optimized image.

The invention can capture the frequently-changed parts in the image by extracting the local dynamic areas of the filtered image, and the parts generally contain important details and dynamic elements, thereby being beneficial to keeping the important information in the image and improving the visual attraction and expressive force of the image. The gray level image conversion of the local dynamic image is beneficial to simplifying image data, reducing the computational complexity and providing more efficient data for subsequent image processing. The image texture extraction is carried out on the local gray level image by using the gray level co-occurrence matrix, so that the texture characteristics of the dynamic image, including blurred dynamic characters or other creatures in the video, can be captured, the visual quality of the image is improved, and the texture sense and detail presentation of the image are enhanced. The gray value calculation formula is used for carrying out gray value calculation on the local gray image, so that the brightness and the contrast of the image can be adjusted to optimize the visual effect of the image, and the visual attractiveness and the definition of the image are improved. The image texture data and the restoration gray value are combined, so that the imperfect part of the local dynamic image can be restored, the overall quality and visual appeal of the image are enhanced, and the image content which is more attractive to users is provided. The filtering image is optimized according to the repairing local dynamic image, so that the details and dynamic effects of the image can be enhanced, the sense of reality and the attraction of the image are improved, the visual experience of a user is improved, and the image is more vivid.

Preferably, the dynamic image gray value calculation formula is as follows:

；

in the method, in the process of the invention,restoration gray value expressed as partial gray image, < +.>Expressed as gray value weights inside the image texture, and>expressed as abscissa and ordinate +.>And->Gray value data within the image texture, when present>Expressed as gray value weights outside the image texture, respectively>Expressed as abscissa and ordinate +.>And->Gray value data which are outside the image texture, < >>The rate of decay of the global gray value expressed as a local gray image,>expressed as gray value mean value inside the image texture, is->Expressed as gray value mean outside the image texture, < >>An abnormal adjustment value expressed as a restored gradation value of the partial gradation image.

The invention utilizes a dynamic image gray value calculation formula which fully considers the gray value weight in the image textureThe abscissa and the ordinate are respectively +.>And->Gray value data +.>Gray value weight outside the image texture>The abscissa and the ordinate are respectively +.>And->Gray value data that is external to the image textureIntegral gray value decay rate of local gray image +. >Gray value mean value inside image texture>Gray value mean value outside the image texture>And interactions between parameters to form a functional relationship:

that is to say,the calculation formula calculates the gray value deviation caused by the inside and the outside of the texture in the dynamic blurred image comprehensively, so that the gray value of the dynamic image part is recalculated, the color is repaired, and the picture with clear color is obtained.

The gray value weight in the image texture is used for adjusting the influence degree of the gray value in the image texture on the repair result. A larger value will enhance details inside the image texture; the abscissa and the ordinate are respectivelyAnd->The gray value data which is positioned in the image texture is one of input values for calculating the repair gray value; the gray value weight outside the image texture is used for adjusting the influence degree of the gray value outside the image texture on the repair result, and the detail outside the image texture is enhanced by a larger value; the abscissa and the ordinate are respectively +.>And->The gray value data that is outside the image texture at the time is one of the input values for calculating the restored gray value. When the intersection point is positioned outside the image texture, the gray value weight outside the image texture is increased, the relative obtained gray value is lower, when the intersection point is positioned inside the image texture, the gray value weight inside the image texture is increased, the relative obtained gray value is higher, the color of the finally obtained image is clearer, and the color of the finally obtained image is better through adjustment >And->The algorithm can protect the detail information inside and outside the image texture, so that the texture characteristics of the original image are kept as much as possible in the repairing process. The integral gray value attenuation rate of the local gray image controls the change speed of the integral gray value, so that the repaired image is smoother in gray. A larger value will result in a faster decay of the overall gray value; the average value of gray values in the image texture is a reference point for normalizing the gray values in the image texture; the average gray value outside the image texture is a reference point for normalizing the gray value outside the image texture so that the algorithm can adapt to the gray value range of different images. According to the functional relation, the dynamic blurred image is effectively restored by weighing gray values of different areas, so that the restored image has higher definition and visual effect, and the restoration effect can be optimized according to specific conditions by adjusting different parameters. Abnormal adjustment value +.>The functional relation is adjusted and corrected, so that the error influence caused by abnormal data or error items is reduced, and the repair gray value of the local gray image is more accurately generated >The accuracy and the reliability of gray value calculation on the local gray image are improved. Meanwhile, the weight information and the adjustment value in the formula can be adjusted according to actual conditions, and the method is applied to pixel points of the local gray level image, so that the flexibility and applicability of the algorithm are improved.

Preferably, step S5 comprises the steps of:

and constructing an automatic forwarding script according to the optimized pre-forwarding data and a preset automatic script frame, automatically and repeatedly forwarding the optimized pre-forwarding data through terminal equipment when the automatic script is executed, generating automatic forwarding content data, and collecting a log file of the automatic forwarding content data to feed back to the terminal.

According to the invention, the automatic forwarding script is constructed according to the optimized pre-forwarding data and the preset automatic script frame, so that the automatic transmission of the content can be realized, manual intervention is not needed, the operation efficiency is improved, and particularly when the content is required to be forwarded on a large scale, the automatic script can greatly save time and manpower resources. The optimized pre-forwarding data is automatically and repeatedly forwarded through the terminal equipment, so that the content presentation mode is guaranteed to accord with the habit of a user and the rule of a platform, the user experience is improved, the content propagation mode is more consistent with the expectations of the user, and the most important forwarding content can be provided at the forefront of a homepage by the repeated forwarding, so that other users are attracted to review. By executing the automation script, the automatic forwarding content data is generated, which is beneficial to ensuring the consistency and accuracy of the content, and the automatically generated content is not easy to be wrong, so that the quality and the credibility of the content are improved. The log file of the automatic forwarding content data is collected and fed back to the terminal, so that the effect of the automatic forwarding process can be monitored and evaluated, potential problems can be identified, an automatic strategy can be improved, and interaction and feedback of a user can be known.

The method and the device have the beneficial effects that by acquiring the authority of the terminal equipment of the user, the real-time monitoring of the instant messaging software is realized, the generation of personalized content and the accurate capture of the interests of the user are improved, the browsing data and the browsing interest level are integrated, a rich data base is provided for the subsequent steps, and personalized recommendation and content optimization are facilitated. The browsing data are divided into the preferred part and the alternative part according to the interests and the browsing interest level of the user, which is beneficial to user experience, so that other users can see the content of the greatest interest of the users, the satisfaction degree and the interactivity are improved, the selection of the users is increased due to the existence of the alternative data, and more diversified contents are provided. By prioritizing the browsing interest levels according to the preferred browsing data, it is ensured that the user first sees the most relevant and interesting content, user satisfaction and interactivity are improved, preferred and alternative data are integrated, and diversity and attractiveness of the content are further enhanced. Through the refinement processing of the text and image contents, the quality and the attraction of the contents are improved, the abstract extraction of the text, the filtering and the dynamic optimization of the images are helpful for users to understand and browse the contents more quickly, the quality of the browsed contents is higher than that of the original works, and the interaction and the retention rate of the users are increased. By constructing and executing the automatic forwarding script, the automatic transmission of the content is realized, the operation efficiency is improved, the time and the resources are saved, the collection of the log file and the feedback to the terminal are beneficial to monitoring and improving the automation strategy, the more attractive content is provided, and the user interaction and satisfaction are improved.

Drawings

Fig. 1 is a schematic flow chart of steps of an automatic forwarding method of a friend circle based on instant messaging software;

FIG. 2 is a detailed flowchart illustrating the implementation of step S1 in FIG. 1;

FIG. 3 is a detailed flowchart illustrating the implementation of step S2 in FIG. 1;

FIG. 4 is a flowchart illustrating the detailed implementation of step S4 in FIG. 1;

the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

In order to achieve the above objective, referring to fig. 1 to 4, the present invention provides an automatic forwarding method for a circle of friends based on instant messaging software, comprising the following steps:

In the embodiment of the present invention, as described with reference to fig. 1, a step flow diagram of an automatic forwarding method for a friend circle based on instant messaging software of the present invention is provided, and in this example, the steps of the automatic forwarding method for a friend circle based on instant messaging software include:

in the embodiment of the invention, the user authorizes the authority of the terminal equipment, and the system is allowed to collect data. When a user browses a friend circle by using instant messaging software, the behavior of the user, such as scrolling, praise or comment, is monitored in real time, and a corresponding browsing data record is generated. When browsing data, the user calculates the time of browsing a piece of friend circle content and the interest degree of the friend circle category so as to calculate the browsing interest level of the user on the current friend circle content, for example, when the user browses a certain theme, the browsing time is longer, and when the user completely views the friend circle content, the corresponding browsing interest level is higher. The browsing data and the browsing interest level are combined to generate browsing interest related data, so that the interests and the favorites of the user can be known more accurately, and powerful data support is provided for subsequent content recommendation and optimization. Not only improves the accuracy of personalized content recommendation, but also is beneficial to improving the user interaction rate and satisfaction, and provides friend circle experience which is more in line with the interests of the user.

in the embodiment of the invention, the browsing interest related data is refined, a series of preset interest thresholds are set based on the browsing interest level of the user, for example, a first interest threshold is set, and when the browsing interest level of the browsing data is higher than the threshold, the data is marked as preferable browsing data, so that the content which is interested in particularly is represented, and the data is highly matched with the current interest of the user. In addition, a second interest threshold is set, and the second interest threshold is smaller than the first interest threshold, when the browsing interest level is between the first interest threshold and the second interest threshold, the first interest threshold is larger than the second interest threshold, the data is marked as alternative browsing data, which means that although the data is related to the interest of the user, the matching degree is not higher than the preferred data, so that two groups of data are generated: the sorting method not only provides sorting of personalized contents, but also provides more diversified content selection for users, thereby improving user interaction and satisfaction.

in the embodiment of the invention, the preferred browsing data is further processed by utilizing the browsing interest level of the user, the preferred browsing data is ordered according to the browsing interest level, and the data which is most in line with the current interest of the user is arranged in front, so that the priority browsing data is formed, for example, the higher the browsing interest level is, the more the preferred browsing data is. The user self-defines and selects a part of alternative browsing data which are related to the interests of the user but do not reach the preferred standard, and then integrates the data with the priority browsing data to generate pre-forwarding data, so that the user is ensured to see the most relevant and attractive contents first, and meanwhile, other relevant contents are provided for maintaining diversity, thereby improving the satisfaction degree and interactivity of the user, and the personalized content sequencing and integration method effectively optimizes the content experience of the friend circle of the user.

In the embodiment of the invention, the text and image content of the pre-forwarding data are comprehensively optimized, key information is extracted and a refined text abstract is generated for the text data, for example, an article contains a large amount of content, the most important part is automatically extracted, and the abstract text is output by using a trained cyclic neural network model, so that the text is ensured to be more concise and attractive, and the understanding and interaction of the text content by a user are facilitated. Meanwhile, for image data, gaussian filtering and other technologies are applied to improve the definition and quality of an image, then local dynamic region extraction and gray image conversion are carried out on the image to extract image textures and repair local gray values, so that the image textures and repair local gray values are used for repairing dynamic blur caused by characters or other dynamic scenes in pictures and videos, for example, moving people in videos are excessively blurred, edges of the characters are extracted, task color main gray values are extracted, secondary drawing and color mapping are carried out to optimize blur caused by a friend circle image, and the steps jointly generate an optimized image, so that the image is attractive and clear.

In the embodiment of the invention, the key process of automatic forwarding is executed, firstly, an automatic forwarding script is constructed based on optimized pre-forwarding data and a preset automatic script frame, the script comprises detailed guidance on how to select, arrange and issue optimized contents, when the automatic forwarding script is executed, the optimized contents are automatically forwarded to a friend circle of a user one by one according to a preset rule to generate automatic forwarding content data, the data comprise forwarded texts, images and other related information, and the contents are automatically generated on the basis of multiple optimization, so that the quality and the attraction of the contents are ensured. Meanwhile, log files of the automatic forwarding content data are correspondingly collected, and detailed information of each forwarding, such as time, content, audience response, whether forwarding is successful or not and the like are recorded. By analyzing the log data, whether the content is successfully transmitted can be known, and further the automation script and the content strategy are continuously improved.

Preferably, step S1 comprises the steps of:

step S11: acquiring terminal equipment authority of a user;

As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S1 in fig. 1 is shown, where step S1 includes:

step S11: acquiring terminal equipment authority of a user;

in the embodiment of the invention, the user is firstly requested to grant the necessary authority of the terminal equipment, such as accessing the camera, the storage, the network and the like. After the user agrees, the required rights are obtained to perform the subsequent operations.

in the embodiment of the invention, the activities of the user on the instant messaging software are monitored in real time according to the obtained authority, including scrolling, praise, comments and the like, for example, the user scrolls a friend circle, the content, the browsing time and the interaction behavior of the user are recorded, and the data are captured in real time and are organized into browsing data.

in the embodiment of the invention, the captured browsing data are subjected to time series analysis to identify the browsing behavior pattern and trend of the user, for example, the content types browsed by the user in different time periods, such as the starting time and the ending time of each friend circle, can be detected, and the time series data are favorable for data integration of the follow-up steps and browsing interest levels generated when the user views the content of the friend circles.

in the embodiment of the invention, the browsing interest level calculation formula is utilized, and the size of the browsing interest level of the user is judged through scrolling, praise, comment, time and the like corresponding to the browsing content in the browsing data, so that whether the user is interested in the browsing content is judged, wherein the higher the browsing interest level is, the more interested the user is in the browsing friend circle content in the browsing data correspondingly.

In the embodiment of the invention, a time sequence data structure is established for each user, wherein each time point corresponds to the browsing activity and the interest level of the user, for example, the time sequence can record the browsing activity and the interest level of the user in the time period by taking each browsing content as a unit. For each time point, the browsing activity and the interest level of the user are recorded in a time sequence, which may be a data record, including information such as browsed content, interest level, time stamp, etc., for example, the user browses the article a within a certain hour, the interest level is 0.7, and the system records the information at the corresponding time sequence point, thereby generating browsing interest related data.

Preferably, the browsing interest level calculation formula is as follows:

；

that is to say,the functional relation calculates the degree of interest of the user in browsing the data content by considering the browsing time, praise number, comment number and the like of the user for a certain piece of friend circle content. The user browsing end time and the starting time represent the browsing time length of the user, and the browsing time is longer to drive the browsing interests of the user; content-to-category ratio in all categories, content-to-categoryThe category of content is rated in the user browsing category and the category of content is rated in the user history forwarding category, meaning that the user has different degrees of interest in different categories of content, e.g., users often browse the content of the technology category and their history forwarding is also dominated by the technology related content, then their interest level may be higher in that category of content. The history number of praise of the content corresponding to the browsing data and the praise number of the content on the same day represent the influence of the praise number, and the praise number can be regarded as an index of the popularity of the content, if a content obtains a large number of praise in the past, the content can be considered to be more attractive, the interest level of the user can be higher, and the praise number on the same day can be used for considering the real-time popularity of the content; the content distribution time corresponding to the browsing data is used to consider the freshness of the content, and for some fields, the newly distributed content may be more attractive to users, so they may get a higher interest level in the calculation. The functional relation aims at accurately calculating the interest level of the user to the specific content according to the browsing behavior of the user, the characteristics of the content and the historical data, so that personalized content recommendation can be better provided for the user, the user experience is improved, and the attraction of the content is improved. Abnormality adjustment value using browsing interest level +. >The functional relation is adjusted and corrected, so that the error influence caused by abnormal data or error items is reduced, and the browsing interest level is generated more accurately>The accuracy and the reliability of interest level calculation of browsing contents on browsing data are improved. Meanwhile, the weight information and the adjustment value in the formula can be adjusted according to actual conditions and are applied to different browsing data for calculation, so that the flexibility and applicability of the algorithm are improved.

Preferably, step S2 comprises the steps of:

As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:

in the embodiment of the invention, interest determination is performed on browsing interest related data to determine whether the browsing interest related data is marked as preferable browsing data, for example, if the browsing interest level of a user is higher than a preset first interest threshold in a specific period of time, it is recognized that the user shows a thick interest in a certain theme or content, for example, if the user is browsing science and technology related friend circle content, by collecting that the user browses the friend circle content for a longer time, the user browses the friend circle content of the category more, etc., the corresponding browsing interest level is 90, and if the first interest threshold is 80, the browsing interest level is greater than the first interest threshold, then the data related to the science and technology will be marked as preferable browsing data.

In the embodiment of the invention, the browsing interest related data with the browsing interest level between the first interest threshold and the second interest threshold is subjected to interest judgment, and is marked as alternative browsing data, for example, when the browsing interest level of the user is between the first interest threshold and the second interest threshold, the first interest threshold is larger than the second interest threshold, if the first interest threshold is 80, the second interest threshold is 50, the user is considered to be interested in related content but not particularly interested in the content, the data may include content related to the interest of the user, for example, the content of a friend who occasionally pays attention, the corresponding browsing interest level is 60, and when the browsing interest level is smaller than the first interest threshold which is 80 and larger than the second interest threshold which is 50, the browsing interest level is larger than the first interest threshold but not sufficiently interested in the user.

In the embodiment of the invention, the browsing interest related data with the browsing interest level lower than the second interest threshold value is removed, for example, the browsing interest level of the user is lower than the second interest threshold value, the interest level of the user in the content is determined to be lower, for example, the content of the friend circle of the list of friends which the user is not usually concerned is scanned, the corresponding browsing interest level is 30 if the content which is scratched at a glance is scanned, and the data is determined not to be contained in the alternative browsing data when the browsing interest level is smaller than the second interest threshold value which is 50.

Preferably, step S3 comprises the steps of:

In the embodiment of the invention, the alternative browsing data is selected according to the personalized preference and the historical behavior of the user in a user-defined manner so as to acquire the content which is most suitable for the user, for example, the user selects the whole content of friends in sports news class or a specific piece of friend circle content, marks the whole content as the selected browsing data, ignores other irrelevant content, and the user-defined selecting process is based on the friend circle content selected by the user. The marked preferred browsing data and the browsing interest level information associated therewith are used to prioritize such data, e.g. an article being browsed by the user and browsing content having a high interest level, as prioritized browsing data. Integrating the priority browsing data with the custom selected alternate browsing data and ordering the alternate browsing data after the priority browsing data to generate pre-forwarding data ensures that the user sees the most attractive and relevant content and can dynamically adjust the content according to the user's interests and behaviors.

Preferably, step S4 comprises the steps of:

As an example of the present invention, referring to fig. 4, a detailed implementation step flow diagram of step S4 in fig. 1 is shown, where step S4 includes:

in the embodiment of the invention, the pre-forwarding data is divided into a text part and an image part, wherein the text part comprises text content and the like, the image part comprises related pictures or videos, and the division process is favorable for the system to better process different types of data and optimize the data.

in the embodiment of the invention, the pre-forwarding text data is analyzed, the key information of the text is extracted, and the text abstract is generated, wherein the text abstract can be a refined version of an article and contains the most important information and key words, so that a user can know the content of the article more quickly, and for a long friend circle content, the text abstract can comprise main events, key characters and background information of the content.

In the embodiment of the invention, the original pre-forwarding text data and the generated text abstract data are integrated to generate the optimized text data, and the optimized text data comprises the original text and the abstract so that other users can select to browse the abstract, and after the abstract is interested, the text content is carefully read so as to attract more users to review the friend circle content.

in the embodiment of the invention, the image part is processed by using a Gaussian filtering technology to reduce noise and improve image quality, and the Gaussian filtering is a common image processing technology, which can smooth the image and reduce detail noise.

in the embodiment of the invention, the image subjected to Gaussian filtering is subjected to local dynamic image optimization, and a plurality of blurring caused by movement of people exist in the image or video, so that the dynamic blurring part of the image is extracted, and corresponding optimization operations are performed, including enhancing the contrast, color saturation and sharpness of the image, so that the image or video looks clearer.

In the embodiment of the invention, the optimized text data and the optimized image data are integrated together to generate the optimized pre-forwarding data, and the optimized text data and the optimized image data are integrated according to the previous sequence, so that the original tedious friend circle content with poor pictures or videos becomes more interesting and the pictures or videos are clearer.

Preferably, step S42 comprises the steps of:

acquiring text-summary data;

The invention can capture the most important and relevant information in the text by extracting the key field of the pre-forwarded text data, thereby improving the accuracy and efficiency of the text abstract, being beneficial to a user to quickly acquire the core content of the article, reducing the risk of information overload and improving the user experience. The method comprises the steps of collecting data related to articles in a cloud server through a web crawler, storing the content and abstract of the articles after data preprocessing, correspondingly integrating the content and abstract, storing the content and abstract in a database, and acquiring text-abstract data in the database, wherein the text-abstract data is a training data set required for building a training abstract prediction model, and is beneficial to learning abstract generation rules from a large number of text examples so as to improve the capability of automatically generating text abstracts. By using deep learning models such as a Recurrent Neural Network (RNN), complex mapping relations between text contents and text summaries can be learned, more accurate and coherent text summaries are generated, and the information integrity of the contents is ensured. By using text-abstract data to carry out model training on an initial abstract prediction model, the performance of the model is continuously improved, the generation quality of the text abstract is improved, abstract content with more grammar correctness and consistency is generated, and the understanding and participation of users on the content are improved. The key text data is transmitted to the abstract prediction model to conduct text abstract prediction, so that a refined text abstract can be generated quickly and automatically, a user can be provided with a more efficient content browsing and understanding mode, and the satisfaction degree of the user is improved, because the user can acquire information more quickly.

In the embodiment of the invention, key information is extracted from pre-forwarded text data, and the information generally comprises main paragraphs, keywords and the like, and is used for subsequent text abstract generation. The method comprises the steps of collecting data related to articles in a cloud server through a web crawler, storing content and abstracts of the articles after data preprocessing, correspondingly integrating the content and abstracts of the articles, storing the data in a database, extracting the text-abstracts data stored in the database after text-abstracts data are obtained, wherein the data pairs comprise original texts and corresponding abstracts, the data pairs are used for training a text abstract generation model and helping the model learn how to extract key information from the texts to generate abstracts, for example, obtaining text content used for training and corresponding text abstracts, and the original texts and abstracts of each article need to be paired for model learning of the abstract generation mode. A cyclic neural network (RNN), which is a neural network suitable for sequence data, is used to build a mapping relationship between text content and text summaries using a model of cyclic neural network (RNN) or other deep learning, where the text sequence can be processed and the relationship between text content and summary encoded into parameters of the model. And (3) performing model training and verification on the initial abstract prediction model by using the text-abstract data, dividing the text-abstract data into a training set and a verification set by using a cross verification mode, giving the training set to the model in the training process, evaluating the model by using the verification set, and adjusting parameters when the evaluating performance is poor, so that the overfitting capacity of the model is reduced, and thus, the model learns the correlation between the text data and the text abstract. The key text data is transmitted to a trained abstract prediction model, which uses the data to generate a corresponding text abstract, which is generated based on a training set of learned text-to-abstract relationships, e.g., the user has entered a piece of key text data, such as a main paragraph of an article, and the model uses the trained parameters to generate the corresponding text abstract, providing a refined version to the user.

Preferably, step S45 comprises the steps of:

In the embodiment of the invention, the region of interest in the filtered image is selected to extract the local dynamic region, the region is extracted by an image segmentation algorithm or manual selection of the region of interest, and the local dynamic image is obtained. The conversion of a colored partial moving image into a gray image is accomplished by standard image processing methods, such as averaging the RGB values of each pixel into one gray value or using a weighted average method to preserve the brightness information of the image, the converted gray image will contain only black and white hues and no color information, thereby generating a partial gray image. For the selected partial Gray image, a Gray-Level Co-occurrence Matrix, GLCM is calculated. The GLCM is a statistical tool for describing image texture, which records the relative positional relationship between different pixel values in an image, and by calculating various statistical features in the GLCM, such as contrast, energy, uniformity, etc., image texture information of a partial gray scale image can be extracted, and these features provide information about image texture complexity, etc., so as to obtain the texture of a dynamic region of the image, such as the outer contour of a moving person, the texture features of the person, etc., and generate image texture data. The gray value of the blurred region is recalculated by using a dynamic image gray value calculation formula, so that the gray value is used for repairing the color of the blurred region of the image, and the calculation formula relates to the self gray value of the blurred image, the gray value generated by ghosting and the like. And remapping character line textures and restoration gray values of image texture data into a picture space to obtain a restored blurred image, converting the restored blurred image into a colored high-definition image according to original conversion conditions because the restored blurred image is a gray image, and reconstructing an image of the original filtered friend circle content according to the colored high-definition image to generate an optimized image.

Preferably, the dynamic image gray value calculation formula is as follows:

；

in the method, in the process of the invention,restoration gray value expressed as partial gray image, < +.>Expressed as gray value weights inside the image texture, and>expressed as abscissa and ordinate +.>And->Gray value data within the image texture, when present>Expressed as gray value weights outside the image texture, respectively>Expressed as abscissa and ordinate +.>And->Gray value data which are outside the image texture, < >>Represented as officeThe overall gray value decay rate of the partial gray image, is->Expressed as gray value mean value inside the image texture, is->Expressed as gray value mean outside the image texture, < >>An abnormal adjustment value expressed as a restored gradation value of the partial gradation image.

The gray value weight in the image texture is used for adjusting the influence degree of the gray value in the image texture on the repair result. A larger value will enhance details inside the image texture; the abscissa and the ordinate are respectivelyAnd->The gray value data which is positioned in the image texture is one of input values for calculating the repair gray value; the gray value weight outside the image texture is used for adjusting the influence degree of the gray value outside the image texture on the repair result, and the detail outside the image texture is enhanced by a larger value; the abscissa and the ordinate are respectively +.>And->The gray value data that is outside the image texture at the time is one of the input values for calculating the restored gray value. When the intersection point is positioned outside the image texture, the gray value weight outside the image texture is increased, the relative obtained gray value is lower, and when the intersection point is positioned inside the image texture, the image is increased The gray value weight in the texture is higher, the relative obtained gray value is higher, the color of the finally obtained image is clearer, and the gray value weight is adjusted>And->The algorithm can protect the detail information inside and outside the image texture, so that the texture characteristics of the original image are kept as much as possible in the repairing process. The integral gray value attenuation rate of the local gray image controls the change speed of the integral gray value, so that the repaired image is smoother in gray. A larger value will result in a faster decay of the overall gray value; the average value of gray values in the image texture is a reference point for normalizing the gray values in the image texture; the average gray value outside the image texture is a reference point for normalizing the gray value outside the image texture so that the algorithm can adapt to the gray value range of different images. According to the functional relation, the dynamic blurred image is effectively restored by weighing gray values of different areas, so that the restored image has higher definition and visual effect, and the restoration effect can be optimized according to specific conditions by adjusting different parameters. Abnormal adjustment value +.>The functional relation is adjusted and corrected, so that the error influence caused by abnormal data or error items is reduced, and the repair gray value of the local gray image is more accurately generated >The accuracy and the reliability of gray value calculation on the local gray image are improved. Meanwhile, the weight information and the adjustment value in the formula can be adjusted according to actual conditions, and the method is applied to pixel points of the local gray level image, so that the flexibility and applicability of the algorithm are improved.

Preferably, step S5 comprises the steps of:

In the embodiment of the invention, the key process of automatic forwarding is executed, firstly, the preset automatic script frame is constructed through python, js and the like based on the optimized preset forwarding data and the preset automatic script frame, an automatic forwarding script is constructed, the script comprises detailed guidance on how to select, arrange and issue optimized contents, when the automatic forwarding script is executed, the optimized contents are automatically and reversely forwarded to friend circles of a user according to preset rules, and because the optimized friend circle content data is sequenced according to the interested degree of the user, if the content is forwarded according to the conventional sequence, the content of most interest can be sequenced to the top through the repeated forwarding, so that the automatic forwarding content data comprising forwarded texts, images and other related information are generated automatically on the basis of multiple optimization, and the quality and attractive force of the content are ensured. Meanwhile, log files of the automatic forwarding content data are correspondingly collected, and detailed information of each forwarding, such as time, content, audience response, whether forwarding is successful or not and the like are recorded. By analyzing the log data, whether the content is successfully transmitted can be known, and further the automation script and the content strategy are continuously improved.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An automatic forwarding method of a friend circle based on instant messaging software is characterized by comprising the following steps:

step S4, including:

step S46: data integration of the corresponding sequence is carried out on the optimized text data and the optimized image, and optimized pre-forwarding data is generated;

2. The automatic forwarding method of friend circle based on instant messaging software according to claim 1, wherein step S1 comprises the steps of:

step S11: acquiring terminal equipment authority of a user;

3. The automatic forwarding method of friend circle based on instant messaging software according to claim 2, wherein the browsing interest level calculation formula is as follows:

；

in the method, in the process of the invention,expressed as browsing interest level->Expressed as user browsing end time->Expressed as user browsing start time->Expressed as the ratio of the category to which the content belongs in all categories,/for>Expressed as a proportion of the category to which the content belongs in the category the user browses >Expressed as the ratio of the category to which the content belongs in the user history forwarding content category, +.>Expressed as browsing time corresponding to browsing data, +.>Content history praise number expressed as correspondence of browsing data,/->The number of praise on the same day indicated as the content corresponding to the browsing data,/->Content distribution time indicated as browsing data correspondence, < >>Expressed as an exponential constant, +.>An anomaly adjustment value expressed as a browsing interest level.

4. The automatic forwarding method of friend circle based on instant messaging software according to claim 1, wherein step S2 comprises the steps of:

5. The automatic forwarding method of friend circle based on instant messaging software according to claim 1, wherein step S3 comprises the steps of:

6. The automated forwarding method of a circle of friends based on instant messaging software according to claim 1, wherein step S42 comprises the steps of:

acquiring text abstract data;

model training is carried out on the initial abstract prediction model by using the text abstract data, and an abstract prediction model is generated;

7. The automated forwarding method of friend circle based on instant messaging software according to claim 1, wherein step S45 comprises the steps of:

8. The automatic forwarding method of friend circle based on instant messaging software according to claim 7, wherein the dynamic image gray value calculation formula is as follows:

；

in the method, in the process of the invention,restoration gray value expressed as partial gray image, < +. >Expressed as gray value weights inside the image texture,expressed as abscissa and ordinate +.>And->Gray value data within the image texture, when present>Expressed as gray value weights outside the image texture, respectively>Expressed as abscissa and ordinate +.>And->Gray value data which are outside the image texture, < >>The rate of decay of the global gray value expressed as a local gray image,>expressed as gray value mean value inside the image texture, is->Expressed as gray value mean outside the image texture, < >>An abnormal adjustment value expressed as a restored gradation value of the partial gradation image.

9. The automatic forwarding method of friend circle based on instant messaging software according to claim 1, wherein step S5 comprises the steps of: